Making Compliance a First-Class Systems Problem in Data Management

Abstract:

As data management moves from isolated databases to geo-distributed and federated ecosystems, compliance has become a central systems challenge. Modern regulations, such as GDPR and CCPA, and governance requirements do not simply determine whether data access is allowed; they constrain how data is moved, disclosed, discovered, shared, and integrated across parties. This shift calls for a broader view of access control: from a mechanism for restricting access inside a database to a foundation for building compliant data ecosystems. In this talk, I present our work, which treats compliance as a first-class systems problem and traces a path from classical access-control questions to compliant data ecosystems.

In the first part of the talk, I will provide an overview of our work spanning multiple layers of this problem. I will show how compliant geo-distributed query processing incorporates data movement and cross-border dataflow restrictions into query optimization; how access control can be extended with query-aware masking and utility-aware disclosure; and how to enable data discovery, sharing, and integration in federated settings without requiring raw data sharing. Together, these works illustrate that compliance should not merely restrict data use but should guide system design so that useful computation and collaboration remain possible within regulatory boundaries.

In the second part, I will focus on our ongoing work on fine-grained access control. Here, the challenge is to efficiently support expressive, content-aware, and parametric data access policies without resorting to expensive conventional query plans. I will present the core ideas behind our approach and discuss how this work advances fine-grained, native support for compliance-aware data management within database systems.

Bio:

Kaustubh Beedkar is an Assistant Professor in the Department of Computer Science and Engineering at the Indian Institute of Technology Delhi and a Junior Fellow at BIFOLD. His research focuses on compliant data management--building systems that enable efficient data access, sharing, and integration under privacy, security, and regulatory constraints. This spans geo-distributed and federated query processing, access control and disclosure mechanisms, and scalable analytics over large and evolving datasets. Kaustubh is the recipient of the ICDE Best Paper Award, the EDBT Best Demonstration Award, and the SIGMOD Distinguished Reviewer Award. He publishes regularly at leading database venues including SIGMOD, VLDB, ICDE, TODS, and EDBT, where he also serves on program committees.